DNA and RNA condensation is of broad interest in both biology and biotechnology. The in vivo packaging of these nucleic acids (NAs) enables the efficient storage of genetic information; strategies for in vitro packaging are crucial for development of therapeutics, e.g., for non-viral gene delivery. Like-charge attraction between DNA duplexes is an essential precursor of DNA condensation and can be induced by multivalent ions with distinctly different physical properties, ranging from the spherical trivalen inorganic ion, cobalt hexa-amine, through the more rod-like, less charge-dense polyamines found in living cell; e.g., spermine and spermidine. Despite an equal importance of RNA to biology, almost nothing is known about double-stranded RNA condensation. The recent discovery of the RNA interference (RNAi) process is strong motivation for determining whether short RNA duplexes can be packaged for therapeutic applications. The broad goal of this proposal is to develop a quantitative mechanism for multivalent ion-NA duplex interactions that leads directly to ion-induced DNA or RNA attraction. This mechanism will have basic physics at its foundation, yet will be detailed enough to address many practical questions such as sequence dependence of nucleic acid condensation and the striking differences in RNA vs. DNA condensation behavior observed experimentally. Existing theoretical models stress the necessity of ion correlations for generation of these attractive forces, yet no consensus exists describing the mechanism for the attraction at atomic level. Current experimental methods alone do not have the resolution to deliver the atomically detailed picture of the diverse nucleic acid condensation phenomena. To understand the mechanism of counterion-induced attraction between DNA or RNA duplexes, we will develop and experimentally test a new approach to modeling ionic environments of nucleic acids that provides the necessary resolution. The approach will integrate grand canonical Monte Carlo simulations based on novel implicit solvent/explicit ion models (for speed and sampling efficiency) with the traditional explicit solvet molecular dynamics simulations on microsecond time scales (for the highest level of detail). The distinctive feature of our approach is that the computational models will be developed in tight integration with the experiment. Predictions of multivalent ion-NA interactions will be validated by experimental measurement of ion association to and spatial distribution around nucleic acids; the data will be used to fine- tune parameters of the model. Small angle X-ray scattering techniques will probe the structure of NA duplex ionic atmospheres, while condensation experiments using UV spectroscopy will provide additional degrees of comparison with specific predictions of the theory. Once agreement between theory and experiment on simpler systems is reached, we will explore the role of nucleic acid sequence and topology in condensation, and develop an atomistic understanding of how the biologically important polyions spermine and spermidine facilitate attraction between nucleic acid fragments.